The agent fulfilled the task by combining limited site-based documentation with external research. The docs site provided only technical capabilities and architecture; business-level information (pricing, competitive positioning) required external searches, indicating the docs are purely developer-focused and lack discovery/evaluation content.

• ›The docs site (steps 1–3) successfully conveyed core product function (cross-channel notification infrastructure, workflow automation, multi-tenancy) but deliberately excluded pricing and competitive analysis—these were the gaps the agent had to fill via web search (steps 7–8).
• ›The agent was forced to type URLs from memory rather than discover them via site navigation (80% of URLs followed from links, but that 20% 'from memory' reflects the site's lack of pricing/comparison pages), indicating the docs treat pricing and positioning as out-of-scope.
• ›Search results (steps 7–8) were high-fidelity sources: they included direct links to knock.app/pricing and competitive comparison pages, which the agent cited in final Sources. These external sources contained the actual pricing tiers ($0–10k free, $250 Starter, custom Enterprise) and competitor comparisons (vs. Braze, Courier) that the docs omitted.
• ›The site's documentation structure (step 5) revealed comprehensiveness on technical topics (workflows, templates, channels, SDKs) but no navigation to business content, confirming the docs are engineer-facing resource materials, not evaluation/sales-funnel content.
• ›The agent successfully synthesized fragmented information: product function from docs, pricing from search-sourced Knock's own pricing page, competitive position from third-party comparison pages (Courier, Sequenzy), demonstrating resourcefulness but also that the primary site under evaluation does not support self-contained task completion.